H/T Jonathan Bagley for the following video of Professor David Spiegelhalter, the next President of the Royal Statistical Society, and a couple of blog posts by him.
It’s a nice little video, making a simple point, and without using any mathematics, all in less than 5 minutes.
On 31st October 2007 the World Cancer Research Fund (WCRF, a charity/umbrella organisation “supporting research into the role of diet and nutrition in the prevention of cancer”) issued a press release to advertise their comprehensive report on the influences of nutrition and physical activity on cancer, “Food, Nutrition, Physical Activity and the Prevention of Cancer: a Global Perspective”.
As well as the actual report and the press release, the WCRF also released 10 “recommendations for cancer prevention”. The report itself was produced by an international panel of experts and consisted of extensive literature reviews of studies on cancer.
One of the findings of the report was that red and processed meat increased the chances of bowel cancer, where one of the specific recommendations made on the press release was that “People should not eat any more than 500g of red meat a week” (original emphasis). In this finding, the report supports conclusions that were arrived at in the previous report from the same organisation, although this time the authors write that the evidence has become even more conclusive. Most news organisations have picked up the story on the 31st or 1st, after several days of prereporting on what “a major new report to be published by the WCRF” is about to conclude.
The precise way the story was reported varied slightly between “Cancer linked directly to obesity” (Channel 4), and“obesity worse for cancer than smoking” (Daily Mail , although this misleading headline was subsequently changed to “Is anything safe to eat? Cancer report adds bacon, ham and drink to danger list”).
The Sun reported the story by singling out “bacon sandwiches” (“Bacon butty cancer risk” ), and the bacon sandwich, probably because it is a stereotypically English comfort food has subsequently become the point of discussion of the report on several other news and comments pieces, for example on the Guardian’s “Comment is Free” site. A lot of the media coverage of the report has focused on the meat section, to the exclusion of the report’s messages on other foods and physical exercise and their positive as well as their negative contributions to cancer.
Reactions to the story in the subsequent commentary (i.e. op-eds and blog comments on newspaper websites) focused almost exclusively on complaints that scientists won’t let us eat anything now, that they’ll discover that everything causes cancer and we may just as well starve to death, or that surely there will soon be a story about how read meat is good for you…
I think the healthists’ method is simply to keep repeating their message – in this case, the bacon-causes-cancer doctrine – until it ceases to meet any resistance. I don’t know whether it got its first outing in 2007, but despite all resistance at that time, here it is again in 2015. And it’ll be back again in another few years. And gradually everyone will believe that bacon sandwiches cause cancer.
The same happened with the smoking-causes-lung-cancer doctrine, first wheeled out in 1950 to intense resistance from the likes of Sir Ronald Fisher, perhaps the foremost statistician of his day. Fisher died in 1964, and so was unavailable to respond to the next iteration of the smoking-causes-lung-cancer dogma. Nor was he available for subsequent iterations. One by one all the critics (and there seem to have been quite a few of them) simply died out. And as resistance weakened, so the general public gradually fell under the spell of the dogma.
In fact, it has often seemed to me that the Doll and Hill British Doctors Study was designed to do exactly this job of message repetition, because it was a prospective study which waited for 50 years for the doctors to gradually die off. But instead of waiting until they were all dead, they published updated figures every 5 or 10 years, which ensured that smoking and cancer would stay in the news, as the media dutifully published their latest numbers. This is a bit like me doing a very slow orbital simulation model run, and reporting the latest position of an asteroid in the solar system on my blog once a week (e.g. “passing quite near Mars today”). It would be of no value to anyone, but it would keep people reminded of my interest in deadly fireballs raining down from the sky, which is what the asteroid will finally do.
It’s actually a propaganda technique. If something is repeated enough times, people will eventually believe it, regardless of whether it’s true or not. And this has been the result with the smoking-causes-lung-cancer dogma: it’s probably the one thing most people know with any certainty.
All this reminds me of another story I seem to remember reading recently – some time in the past month or two – that scans for breast cancer were actually pretty useless, and threw up false positives as often as true ones. Am I right about this? Anyway, it reminded me that I’d seen the story two years earlier, and written about it on my blog on two occasions.
1% of women at age forty who participate in routine screening have breast cancer. 80% of women with breast cancer will get positive mammographies. 7.273% of women without breast cancer will also get positive mammographies. A woman in this age group had a positive mammography in a routine screening. What is the probability that she actually has breast cancer?
And in fact, it was a problem that I’d first encountered several years before that as a question in a statistical essay. I had assumed back then that it was just an imaginary example dreamt up by some statistician to illustrate Bayes’ Theorem. I only remember it because I had to concentrate very, very hard to do the mathematics (Newtonian mechanics is much easier than probability theory, in my experience), and was rather elated to get the right answer in the end. But I’m now beginning to think that the futility of breast scans was actually known about 10 or more years ago, and is only now being accepted. It hadn’t been dreamt up by a statistics teacher. It was a true story all along. But it needed repeating.
P.S. I’ve just had a reply from Professor Peter Diggle, the current President of the Royal Statistical Society. It begins: “I completely agree with you.”
Hurrah! Hurrah! And again hurrah!!!